The First-Mile Industrial Data Plane
An industrial data plane is the layer where data from physical systems is acquired, structured, and prepared before it reaches any downstream system. This layer does not exist in traditional industrial architectures. KŌJŌ Stack establishes it at the first mile.
System Overview
One stack, from sensor to cloud
Industrial data moves across four secured zones-OT, the customer DMZ, your cloud, and the KŌJŌ Stack control plane. Every hop is authenticated and every connection is initiated from the inside out.
The Data Boundary
Industrial data architectures are divided into two domains
On one side: operational systems-PLCs, sensors, SCADA, and historians. These generate raw, protocol-bound data at high frequency, with no inherent structure beyond device-level addressing.
On the other: enterprise and cloud systems-analytics platforms, AI/ML pipelines, data lakes, and business applications. These require structured, contextualized, reliable data to function.
KŌJŌ Stack is the data boundary between them. It is the layer where data is acquired, structured, filtered, and routed-where raw telemetry becomes usable. This is the industrial data plane. Without it, every downstream system must independently reconstruct data from raw, protocol-bound sources.
The architectural position
OT Domain
PLCs · Sensors · SCADA · Historians
Raw, protocol-bound, unstructured
Data Boundary
KŌJŌ Stack - First-Mile Data Plane
Ingest · Structure · Condition · Route
IT / Cloud Domain
Analytics · AI/ML · Data Lakes · Apps
Structured, contextualized, reliable
Left Side of the Boundary
OT Systems: Where Data Originates
Industrial data originates at PLCs, sensors, SCADA systems, and historians-each speaking different protocols with different timing models, data formats, and addressing schemes. Data here is raw, unstructured, and protocol-bound.
OPC UA
Subscriptions, Browse, security policies
OPC DA
Legacy SCADA / DCS, quality code preservation
Modbus TCP/RTU
Register polling, coils, discrete inputs
Siemens S7
Native protocol, no OPC server required
BACnet IP
COV subscriptions, polling modes
EtherNet/IP (CIP)
Allen-Bradley, Rockwell ecosystems
Sparkplug B
100% spec-compliant MQTT industrial standard
DNP3
Utility SCADA, polling + unsolicited modes
MQTT / Kafka
Broker ingress, topic filtering
The problem: each protocol has its own transport, timing model, data representation, and addressing. Without a structuring layer, every downstream system must independently solve protocol translation, context mapping, and reliability-leading to fragmented, inconsistent, and brittle architectures.
The Data Plane
KŌJŌ Stack: Where Data Becomes Usable
The industrial data plane is where raw telemetry becomes structured, contextualized, and reliable. KŌJŌ Stack owns twelve responsibilities that define what data exists downstream and how it behaves. Every responsibility executes at the edge, with deterministic behavior and bounded latency.
Downstream systems do not define data structures-KŌJŌ Stack does.
- Data is acquired here-protocol-native ingestion at deterministic intervals
- Data is structured here-ISA-95 namespace, canonical schema, provenance metadata
- Data is contextualized here-identity, timestamp, quality, and source context
- Data is prepared here-filtered, normalized, and routed for downstream consumption
This is the layer where industrial data becomes usable.
Data Acquisition
Protocol-native ingestion at deterministic intervals or server-push subscriptions. Each adapter speaks the native language of the device-no translation gateways. Timestamps, quality indicators, and device metadata are preserved at the point of origin.
Data Transformation
Normalization, scaling, unit conversion, and enrichment using the CEL expression engine. Report-by-Exception (RBE) deadband filtering reduces data volume by 90%+ while preserving all meaningful state transitions. Transforms are pure functions-same input, same output, always.
Contextualization
Every data point is mapped into an ISA-95 compliant Unified Namespace hierarchy: Enterprise → Site → Area → Line → Cell. Tags carry identity, timestamp, quality, and source context. The namespace is the contract between all producers and consumers.
Pipeline Execution
Event-driven, deterministic pipelines with bounded latency. Data is structured and prepared before downstream systems consume it-triggered by state changes, not arbitrary polling intervals. Events are delivered in a consistent, deterministic sequence within local pipeline execution paths. Behavior is reproducible and auditable.
Buffering & Reliability
Durable local buffering maintains data continuity during network interruptions. Data is persisted before acknowledgment. On reconnection, buffered data replays in order with original timestamps preserved. Delivery guarantees are a function of the buffering and replay mechanism, not external SLAs.
Extensibility (Module Control Plane)
External modules communicate with the core runtime over typed, versioned interfaces. Each module has independent lifecycle: deploy, start, stop, update, and remove without affecting other components. OEMs and developers extend the data plane with new protocol adapters, transforms, or connectors-without modifying core code.
SDK for Extensibility
The SDK enables developers to extend the data plane to any system-building custom source adapters, destination connectors, and processing modules that integrate natively with the runtime. Partners and OEMs use the SDK to embed KŌJŌ Stack capabilities into their own products. The SDK is the extensibility layer of the data plane.
Co-Located Edge Execution
Workloads execute alongside data pipelines in a shared runtime, ensuring processing occurs where data is acquired and structured. AI inference, protocol adapters, analytics engines, and custom logic operate as managed workloads with lifecycle control, health checks, and secret injection-co-located with the data plane at the edge.
Observability
Per-pipeline throughput, latency (p50/p95/p99), error rates, and backpressure metrics. Module health monitoring with connection status, buffer depth, and protocol diagnostics. Hash-chained audit trail for compliance. Operational control requires operational visibility.
API-First Control
120+ REST API endpoints with OpenAPI 3.0 specification and interactive Swagger UI. Every platform capability - sources, pipelines, destinations, modules, workloads, secrets, certificates - is API-accessible. Designed for automation, CI/CD pipelines, and infrastructure-as-code workflows.
Fleet Management
Centralized deployment, orchestration, and monitoring of all edge nodes across sites, regions, and environments. Push configuration updates, namespace models, and pipeline definitions with controlled rollout. Aggregate telemetry and health across the entire fleet from a single control interface.
AI Agent Integration (MCP Server)
A fleet-aware MCP (Model Context Protocol) server deployed at the edge enables AI agents to discover, query, and diagnose edge deployments. OAuth 2.0 identity with three-axis scoping, 22 read-only tools, and team-level isolation - designed for safe autonomous operation across distributed edge infrastructure.
Right Side of the Boundary
Downstream Systems: Where Data Is Consumed
Cloud platforms, historians, data lakes, AI/ML systems, and enterprise applications depend on structured, reliable data produced by the first-mile data plane. KŌJŌ Stack determines what data these systems receive-and in what form.
S3 Tables / Iceberg
Lakehouse with time travel
Apache Kafka
Event streaming, SASL/TLS
TimescaleDB
Time-series historian
MQTT Brokers
Cloud IoT, QoS 0/1/2
AWS IoT Core
X.509 certificate auth
S3 Data Lake
JSONL/CSV/Parquet batch export
Google Cloud Storage
JSONL/CSV/Parquet, BigQuery compatible
InfluxDB
Time-series historian, Line Protocol
The result: every downstream system receives data that is already structured, contextualized, and quality-annotated. No ETL pipelines to reconstruct meaning. No data quality issues to debug. No protocol-specific logic in analytics code.
How Data Moves
Data is not piped-it is shaped
KŌJŌ Stack does not passively transport data from source to destination. It actively determines what data exists downstream by ingesting, structuring, processing, and distributing it. Data is structured and prepared before downstream systems consume it.
Ingested
Protocol-native acquisition from OT systems
Structured
ISA-95 contextualization and schema normalization
Processed
CEL transforms, RBE filtering, quality annotation
Distributed
Buffered, reliable delivery to every destination
KŌJŌ Stack determines what data exists downstream.
No downstream system sees data that the data plane has not explicitly acquired, structured, and routed. This is first-mile data ownership.
Architectural Position
Where data is shaped determines how data behaves
KŌJŌ Stack does not sit beside existing systems-it sits between them. Every system above this boundary receives data that has been explicitly acquired, structured, and prepared. Every system below it remains unchanged.
The architecture is the product.
One deployment boundary. One data plane. One place where industrial data is acquired, structured, and prepared before it enters the digital world.